Ongoing Technical Research Evaluation of RL for safety and trustworthiness

Evaluating RL agents for safe and trustworthy deployment in critical infrastructure

Overview

Evaluating black-box RL agents deployed in telecom settings (e.g., antenna tilt/intent optimization) for KPI coverage, safety-constraint violations, and robustness to noise/distribution shift; building a certification pipeline for RL agent guardrails, in collaboration with Ericsson.

Related projects

LLM-based Fine Tuning of Restless Multi-armed Bandits for Public Health – Fairness in Multilingual Settings
LLM-based Fine Tuning of Restless Multi-armed Bandits for Public Health – Fairness in Multilingual Settings
healthcare ai-and-society
Evaluating Small Language Models for Detecting Media Bias through Fine-tuning
Evaluating Small Language Models for Detecting Media Bias through Fine-tuning
ai-and-society lm
A Causal Approach for Unfair Edge Prioritization and Discrimination Removal
A Causal Approach for Unfair Edge Prioritization and Discrimination Removal
ai-and-society